FedViT: Federated continual learning of vision transformer at edge

Deep Neural Networks (DNNs) have been ubiquitously adopted in internet of things and are becoming an integral part of our daily life. When tackling the evolving learning tasks in real world, such as classifying different types of objects, DNNs face the challenge to continually retrain themselves according to the tasks on different edge...

Continual prune-and-select: Class-incremental learning with specialized subnetworks

The human brain is capable of learning tasks sequentially mostly without forgetting. However, deep neural networks (DNNs) suffer from catastrophic forgetting when learning one task after another. We address this challenge considering a class-incremental learning scenario where the DNN sees test data without knowing the task from which this...

Dynamic Incremental Learning for real-time disturbance event classification

With recent telemetric advancements, the real-time availability of power grid measurements has opened challenging opportunities for the design of advanced protection and control schemes. Artificial neural networks (ANN) are promising approaches for detecting and classifying disturbance events from measurement data. Numerous offline ANN-based...

Salam Emergency Station: Responding to ermergency

It’s August 4 when Lebanon’s capital, Beirut, is hit by one of the largest non-nuclear explosions the world has ever seen. Unfortunately, this explosion that damaged and/or demolished more than half of the city is not the first crisis it has to deal with. Several catastrophes have already plagued Lebanon and its capital during the last century...

Reliability of LED-based systems

Reliability is an essential scientific and technological domain intrinsically linked with system integration. Nowadays, semiconductor industries are confronted with ever-increasing design complexity, dramatically decreasing design margins, increasing chances for and consequences of failures, shortening of product development and qualification...

Evaluating catastrophic forgetting of state-of-the-art NLP models for predicting moral values

Personal moral values represent the motivation behind individuals' actions and opinions. Understanding these values is helpful both in predicting individuals' actions, such as violent protests, and building AI that can better collaborate with humans. Predicting moral values is a challenging problem due to the abstract and subjective essence of...

The Precipice – Existential Risk and the Future of Humanity. Ord, T., 2020 London, Bloomsbury Publishing. 480 pp, £22.50 (hd)

Comprehensive evaluation of water resources utilization benefits based on catastrophe theory

The benefits of water resource utilization include social benefits, economic benefits, and ecological environmental benefits, so it is difficult to evaluate them comprehensively by adopting the traditional cost-income analysis method. Catastrophe evaluation method, which majors in dealing with uncertain problems, evaluates system...

Health Monitoring for Lighting Applications

Traditional lighting is focused on the prevention of hardware failures. With the trend towards controlled and connected systems, other components will start playing an equal role in the reliability of it. Here reliability needs to be replaced by availability and other modeling approaches are to be considered. System prognostics and health...

A Reliability Prediction Methodology for LED Arrays

In this paper, a physics of failure-based prediction method is combined with statistical models to consider the impact of current crowding and current droop effects on the reliability of LED arrays. Electronic-thermal models of LEDs are utilized to obtain the operation conditions under the influences of current crowding and current droop. A...

A method for fast evaluation of potential consequences of dam breach

Dam breach has catastrophic consequences for human lives and economy. In previous studies, empirical models are often, to a limited extent, due to the inadequacy of historical dam breach events. Physical models, which focus on simulating human behavior during floods, are not suitable for fast analysis of a large number of dams due to the...

Dashain, new civic places for a society in transition

On September 16th 2015, Nepal signed its new constitution, which, among others articles, claims the federalization of the country, makes the nation secular and allows freedom of expression. This event represents a milestone and a glimpse of hope after decades of internal political instability and uncertain administrations, which even led to ten...

A Reliability Prediction for Integrated LED Lamp with Electrolytic Capacitor-Free Driver

This paper studies the interaction of catastrophic failure of the driver and LED luminous flux decay for an integrated LED lamp with an electrolytic capacitor-free LED driver. Electronic thermal simulations are utilized to obtain the lamp's dynamic history of temperature and current for two distinct operation modes: constant current mode (CCM...

Resilience to flooding: Draft building code

A significant issue associated Flood Defence Systems (FDS) is the difficulty of predicting how these structures will behave when inevitably they have been loaded beyond their designed capacity by a flood. The flood can cause these structures to fail catastrophically with loss of life and substantial damage to property. For a limited incremental...

Bankruptcy by catastrophes for major multi-nationals: Stock exchange sensitivity for three catastrophes

This paper investigates the effect of major catastrophes have on stock exchange values for the major multi-nationals. The paper demonstrates that the Sharpe analysis is more sensitive in identifying effects than just following the daily stock values for assessing market response. It was found that major multi-nationals are capable of absorbing...

How safe is safe enough?

Disasters can never be completely ruled out. The Dutch national government has therefore committed itself to the concept of risk rather than the false promise of absolute safety. The objectives of this study were to evaluate current regulatory practices in the domains of industrial and flood safety in the Netherlands, and to formulate proposals...